I will link the potential voting and final voting to a few events that occurred as the contest unfolded over time. To do this, I will use a respected and sophisticated poll that turned out to be correct, the USC Dornsife/Los Angeles Times “Daybreak” poll; or see here for a more detailed graph that marks some events on the timeline. I will not use a prediction market because their predictions turned out to be correct. I cannot reproduce the poll graph here; you will will have to link to the article to see it.
The gray area shows the area of uncertainty or margin of error. Readings within this area denote “too close to call”. The red line is Trump, and the blue line is Clinton.
The single event nearest to the election that turned the election in Trump’s favor and against Clinton was the Oct. 28 e-mail announcement made by FBI Director James Comey. On Oct. 27, the election was a dead heat (Trump, 45.2%; Clinton 44.5%). By November 1, the contest had moved out of the gray area (Trump, 46.9%; Clinton 43.3%). Trump maintained that lead to the end; his best reading was on November 6 (Trump, 48.2%; Clinton, 42.6%).
Comey had earlier cleared Clinton on July 5. The Oct. 28 announcement was a surprise. On that day and thereafter, details emerged. The FBI had seized a computer shared by Huma Abedin with her estranged husband, Anthony Weiner, who was under investigation on sexting charges. The computer, which was insecure, contained 650,000 e-mails. Many were to and from Abedin, Clinton’s close aide. It was reported that Abedin had used the laptop computer to send thousands of e-mails to Clinton. On Sunday November 6, Comey cleared Clinton for a second time. On November 8, Trump stood at 46.8% and Clinton at 43.6%. This is the final poll reading, and it correctly called the election.
The methodology I am using here imitates in a rough way the “event study” methods used in the finance literature to assess the impact of various events on stock prices. Looking back over the data supplied by this poll, certain signal events can be seen clearly. The two conventions each caused a bump upwards for the two candidates. Trump rose about 5% (from about 42% to 47%) when the Republicans convened. Clinton also rose about 5% when the Democrats had their convention. After that Trump went into a slump, then recovered.
At the eve of Clinton’s pneumonia revelation, the two candidates were running neck and neck. The race was too close to call. On September 11, Clinton was at 45.0%, Trump at 43.6%. Clinton’s health became an issue on Sept. 12. Trump went out to a significant lead at that point. One week later (Sept. 19), Trump was at 47.8% and Clinton at 41.1%.
The first two debates were important events. They are associated with Clinton regaining lost ground and Trump losing his lead. By October 12, the race was a dead heat. The third debate appears to have had little impact.
The Clinton campaign handled the pneumonia issue badly. This hurt Clinton’s credibility.
I can’t assess the impact of the initial e-mail scandal that ended on July 5 because the poll starts on July 11. At that point, Trump measured 42.7% and Clinton 40.3%.
The October 28 announcement focused attention on Abedin, Weiner and Clinton, linking the three of them and raising all sorts of questions. They include lax security, incompetent handling of government affairs and possible hearings and indictments. The spotlight also fell on Clinton’s very close relationship to Abedin. We do not know what went through the minds of those polled, but we do know that Comey’s announcement had a marked negative effect on Clinton’s numbers.
This attribution has problems, however. It’s not definitive. There are two reasons for some doubt. The major one is that in the time frame that covers the 10 days before the election, we expect that undecided people will make up their minds. Voting preferences will crystallize in this swing group. There will be some movement in one direction or another; the question is how the undecideds broke and how the Comey announcement affected their votes. We do not know. The second reason is that a movement toward Trump appears to have begun on October 24 and carried to October 26. Trump rose from 43.8% to 45.3% and took the lead from Clinton, who fell from 45.1% to 44.2%. These are small changes and within polling error, but they are still visible.
Research begets the call for more research. But I do not see how we can reach any other conclusion than that the Abedin-Weiner-Clinton factor was sparked by FBI Director James Comey. He lit the blaze on October 28 that consumed the Clinton candidacy and his November 6 announcement was too late to extinguish it.
9:00 am on November 12, 2016
